hard20 pts · first answer countsdeep-learningpytorchautograd

Error from an in-place operation

backward() raises "one of the variables needed for gradient computation has been modified by an inplace operation". What usually causes it?

A.A tensor saved for backward was changed in placeB.The loss was a float instead of a tensorC.The model is on the GPUD.zero_grad() was called before backward

Discussion 0

Join free to post in the discussion.

Error from an in-place operation

Discussion 0

Related questions